import pandas as pd

df = pd.read_csv(
    'data\AirQualityUCI.csv',
    sep=';',
    decimal=',',
)
df.head().iloc[:,11:17]

import numpy as np
df = df.drop(columns=["Unnamed: 15", "Unnamed: 16"])
df.replace(-200.0, np.nan, inplace=True)
df = df[:-114]

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9357 entries, 0 to 9356
Data columns (total 15 columns):
 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   Date           9357 non-null   object 
 1   Time           9357 non-null   object 
 2   CO(GT)         7674 non-null   float64
 3   PT08.S1(CO)    8991 non-null   float64
 4   NMHC(GT)       914 non-null    float64
 5   C6H6(GT)       8991 non-null   float64
 6   PT08.S2(NMHC)  8991 non-null   float64
 7   NOx(GT)        7718 non-null   float64
 8   PT08.S3(NOx)   8991 non-null   float64
 9   NO2(GT)        7715 non-null   float64
 10  PT08.S4(NO2)   8991 non-null   float64
 11  PT08.S5(O3)    8991 non-null   float64
 12  T              8991 non-null   float64
 13  RH             8991 non-null   float64
 14  AH             8991 non-null   float64
dtypes: float64(13), object(2)
memory usage: 1.1+ MB

df.shape

(9357, 15)

import pandas as pd

df = pd.read_csv(
    'data\AirQualityUCI.csv',
    sep=';',
    decimal=',',
)
df.head().iloc[:,11:17]
df1=df.copy()

import numpy as np
df = df.drop(columns=["Unnamed: 15", "Unnamed: 16", "Date", "Time"])
df.replace(-200.0, np.nan, inplace=True)
df = df[:-114]

df=df.dropna()

import pandas as pd
import numpy as np
import scipy.stats as st
import seaborn as sns

arr = df['PT08.S3(NOx)']

import matplotlib.pyplot as plt

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr, kde=True, bins=30)
plt.title("Histogram: PT08.S3(NOx)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr)
plt.title("Boxplot: PT08.S3(NOx)")
plt.xlabel("PT08.S3(NOx)")

plt.tight_layout()
plt.show()

arr1 = df['NO2(GT)']
plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: NO2(GT)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: NO2(GT)")
plt.xlabel("NO2(GT)")

plt.tight_layout()
plt.show()

import numpy as np
import scipy.stats as st
import matplotlib.pyplot as plt
import seaborn as sns

arr2 = df['C6H6(GT)']
plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr2, kde=True, bins=30)
plt.title("Histogram: C6H6(GT)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr2)
plt.title("Boxplot: C6H6(GT)")
plt.xlabel("C6H6(GT)")

plt.tight_layout()
plt.show()

arr1 = df['CO(GT)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: CO(GT)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: CO(GT)")
plt.xlabel("CO(GT)")

plt.tight_layout()
plt.show()

arr1 = df['NMHC(GT)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: NMHC(GT)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: NMHC(GT)")
plt.xlabel("NMHC(GT)")

plt.tight_layout()
plt.show()

arr1 = df['PT08.S2(NMHC)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: PT08.S2(NMHC)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: PT08.S2(NMHC)")
plt.xlabel("PT08.S2(NMHC)")

plt.tight_layout()
plt.show()

arr1 = df['PT08.S1(CO)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: PT08.S1(CO)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: PT08.S1(CO)")
plt.xlabel("PT08.S1(CO)")

plt.tight_layout()
plt.show()

arr1 = df['NOx(GT)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: NOx(GT)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: NOx(GT)")
plt.xlabel("NOx(GT)")

plt.tight_layout()
plt.show()

arr1 = df['PT08.S4(NO2)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: PT08.S4(NO2)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: PT08.S4(NO2)")
plt.xlabel("PT08.S4(NO2)")

plt.tight_layout()
plt.show()

arr1 = df['PT08.S5(O3)']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: PT08.S5(O3)")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: PT08.S5(O3)")
plt.xlabel("PT08.S5(O3)")

plt.tight_layout()
plt.show()

arr1 = df['AH']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: AH")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: AH")
plt.xlabel("AH")

plt.tight_layout()
plt.show()

arr1 = df['T']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: T")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: T")
plt.xlabel("T")

plt.tight_layout()
plt.show()

arr1 = df['RH']

plt.figure(figsize=(14,6))
plt.subplot(1,2,1)
sns.histplot(arr1, kde=True, bins=30)
plt.title("Histogram: RH")
plt.xlabel("Wartość")
plt.ylabel("Częstość")

plt.subplot(1,2,2)
sns.boxplot(x=arr1)
plt.title("Boxplot: RH")
plt.xlabel("RH")

plt.tight_layout()
plt.show()

import matplotlib.pyplot as plt
import seaborn as sns


corr = df.corr()


plt.figure(figsize=(12,10))
sns.heatmap(corr, 
            annot=True,        
            fmt=".2f",       
            cmap="coolwarm",   
            cbar_kws={"shrink": .8})
plt.title("Mapa cieplna korelacji między wszystkimi zmiennymi", fontsize=16)
plt.xticks(rotation=45, ha='right')
plt.yticks(rotation=0)
plt.tight_layout()
plt.show()

from pandas.plotting import scatter_matrix
import matplotlib.pyplot as plt


scatter_matrix(df, 
               alpha=0.5, 
               diagonal='hist',
               figsize=(12,12),
               s=20,            
               hist_kwds={'bins': 20, 'edgecolor':'k'})
plt.suptitle("", y=1.02, fontsize=16)
plt.tight_layout()
plt.show()

df1= df1.drop(columns=["Unnamed: 15", "Unnamed: 16"])
df1.replace(-200.0, np.nan, inplace=True)
df1= df1[:-114]

import pandas as pd
import matplotlib.pyplot as plt


df1['Time_clean'] = df1['Time'].str.replace('.', ':', regex=False)


df1['Datetime'] = pd.to_datetime(
    df1['Date'] + ' ' + df1['Time_clean'],
    format='%d/%m/%Y %H:%M:%S'
)
df1 = df1.set_index('Datetime')

df1 = df1.drop(columns=['Date', 'Time', 'Time_clean'])

import matplotlib.pyplot as plt

cols = df1.columns.tolist()
n = len(cols)
ncols = 2
nrows = (n + ncols - 1) // ncols


axes = df1[cols].plot(
    subplots=True,
    layout=(nrows, ncols),
    figsize=(12, nrows * 3),
    sharex=False
)


for ax, col in zip(axes.flatten(), cols):
    ax.set_xlabel("Czas")                 
    ax.set_ylabel("Wartość pomiaru")       
    ax.set_title(col)                      

if nrows * ncols > n:
    for ax in axes.flatten()[n:]:
        fig = ax.get_figure()
        fig.delaxes(ax)

plt.tight_layout()
plt.show()

import pandas as pd
import numpy as np

df = pd.read_csv("data/AirQualityUCI.csv", sep=";", decimal=",")
df

df = df[:-114]
df = df.drop(columns=["Time", "Date", "Unnamed: 15", "Unnamed: 16"])
df.replace(-200.0, np.nan, inplace=True)
print("Data shape:", df.shape)

Data shape: (9357, 13)

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9357 entries, 0 to 9356
Data columns (total 13 columns):
 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   CO(GT)         7674 non-null   float64
 1   PT08.S1(CO)    8991 non-null   float64
 2   NMHC(GT)       914 non-null    float64
 3   C6H6(GT)       8991 non-null   float64
 4   PT08.S2(NMHC)  8991 non-null   float64
 5   NOx(GT)        7718 non-null   float64
 6   PT08.S3(NOx)   8991 non-null   float64
 7   NO2(GT)        7715 non-null   float64
 8   PT08.S4(NO2)   8991 non-null   float64
 9   PT08.S5(O3)    8991 non-null   float64
 10  T              8991 non-null   float64
 11  RH             8991 non-null   float64
 12  AH             8991 non-null   float64
dtypes: float64(13)
memory usage: 950.4 KB

print("Liczba zduplikowanych wierszy:", df.duplicated().sum())
df.drop_duplicates(inplace=True)

Liczba zduplikowanych wierszy: 31

df.isna().sum()

CO(GT)           1653
PT08.S1(CO)       335
NMHC(GT)         8412
C6H6(GT)          335
PT08.S2(NMHC)     335
NOx(GT)          1608
PT08.S3(NOx)      335
NO2(GT)          1611
PT08.S4(NO2)      335
PT08.S5(O3)       335
T                 335
RH                335
AH                335
dtype: int64

df = df.drop("NMHC(GT)", axis=1)

df[df["PT08.S1(CO)"].isna()].drop(["CO(GT)", "NOx(GT)", "NO2(GT)"], axis = 1)

df = df.drop(
    df[df["PT08.S1(CO)"].isna()].drop(["CO(GT)", "NOx(GT)", "NO2(GT)"], axis = 1).index
    ).reset_index(drop=True)

from sklearn.model_selection import train_test_split

X = df.drop("C6H6(GT)", axis=1)
y = df["C6H6(GT)"]

X_train, X_temp, y_train, y_temp = train_test_split(X, y, test_size=0.2, random_state=42)
X_val, X_test, y_val, y_test = train_test_split(X_temp, y_temp, test_size=0.5, random_state=42)

X_train.shape, X_val.shape, X_test.shape

((7192, 11), (899, 11), (900, 11))

from sklearn.impute import KNNImputer

imputer = KNNImputer()

X_train_imputed = pd.DataFrame(imputer.fit_transform(X_train), columns=X_train.columns, index=X_train.index)
X_val_imputed = pd.DataFrame(imputer.transform(X_val), columns=X_val.columns, index=X_val.index)
X_test_imputed = pd.DataFrame(imputer.transform(X_test), columns=X_test.columns, index=X_test.index)

X_train_imputed.isna().sum().sum(), X_val_imputed.isna().sum().sum(), X_test_imputed.isna().sum().sum()

(0, 0, 0)

import math
import matplotlib.pyplot as plt
import seaborn as sns

cols = X_train_imputed.columns
n = len(cols)
cols_per_row = 3
rows = math.ceil(n / cols_per_row)

fig, axes = plt.subplots(rows, cols_per_row, figsize=(18, rows * 4))
axes = axes.flatten()

for i, col in enumerate(cols):
    sns.boxplot(x=X_train_imputed[col], ax=axes[i], color="skyblue")
    axes[i].set_title(col)

for j in range(i + 1, len(axes)):
    axes[j].set_visible(False)

plt.tight_layout()
plt.show()

numeric_cols = X_train_imputed.select_dtypes(include=[np.number])

Q1 = numeric_cols.quantile(0.25)
Q3 = numeric_cols.quantile(0.75)
IQR = Q3 - Q1

is_outlier = (numeric_cols < (Q1 - 1.5 * IQR)) | (numeric_cols > (Q3 + 1.5 * IQR))

outliers_per_column = is_outlier.sum()
print("Liczba outlierów w zbiorze treningowym:\n\n", outliers_per_column)

total_outliers = is_outlier.sum().sum()
rows_w_outliers = len(X_train_imputed[is_outlier.any(axis=1)])

print(f"\nŁącznie {total_outliers} wartości odstających w {rows_w_outliers} wierszach.")

Liczba outlierów w zbiorze treningowym:

 CO(GT)           248
PT08.S1(CO)      100
PT08.S2(NMHC)     58
NOx(GT)          399
PT08.S3(NOx)     195
NO2(GT)           75
PT08.S4(NO2)      83
PT08.S5(O3)       69
T                  3
RH                 0
AH                 1
dtype: int64

Łącznie 1231 wartości odstających w 735 wierszach.

X_train_clipped = X_train_imputed.copy()

for col in numeric_cols.columns:
    Q1 = X_train_imputed[col].quantile(0.25)
    Q3 = X_train_imputed[col].quantile(0.75)
    IQR = Q3 - Q1
    lower_bound = Q1 - 1.5 * IQR
    upper_bound = Q3 + 1.5 * IQR

    X_train_clipped[col] = np.clip(X_train_imputed[col], lower_bound, upper_bound)


outliers = ~is_outlier.any(axis=1)
X_train_removed_outliers = X_train_imputed[outliers]
y_train_removed_outliers = y_train[outliers]

def build_and_evaluate(X_train, y_train, X_val, y_val):

    from xgboost import XGBRegressor
    from sklearn.linear_model import LinearRegression
    from sklearn.metrics import mean_squared_error

    model = XGBRegressor()
    model.fit(X_train, y_train)
    y_pred = model.predict(X_val)
    rmse_xgb = np.sqrt(mean_squared_error(y_val, y_pred))

    model = LinearRegression()
    model.fit(X_train, y_train)
    y_pred = model.predict(X_val)
    rmse_lr = np.sqrt(mean_squared_error(y_val, y_pred))

    return rmse_xgb, rmse_lr

rmse_imputed, rmse_imputed_lr = build_and_evaluate(X_train_imputed, y_train, X_val_imputed, y_val)
rmse_clipped, rmse_clipped_lr = build_and_evaluate(X_train_clipped, y_train, X_val_imputed, y_val)
rmse_removed_outliers, rmse_removed_outliers_lr = build_and_evaluate(X_train_removed_outliers, y_train_removed_outliers, X_val_imputed, y_val)

print("XGBoost RMSE (odporny na outliers):")
print(f"RMSE for data with outliers: {rmse_imputed:.2f}")
print(f"RMSE for clipped outliers: {rmse_clipped:.2f}")
print(f"RMSE for removed outliers: {rmse_removed_outliers:.2f}")

print("\nLinear Regression RMSE (wrażliwy na outliers):")
print(f"RMSE for data with outliers: {rmse_imputed_lr:.2f}")
print(f"RMSE for clipped outliers: {rmse_clipped_lr:.2f}")
print(f"RMSE for removed outliers: {rmse_removed_outliers_lr:.2f}")

XGBoost RMSE (odporny na outliers):
RMSE for data with outliers: 0.35
RMSE for clipped outliers: 0.45
RMSE for removed outliers: 1.36

Linear Regression RMSE (wrażliwy na outliers):
RMSE for data with outliers: 1.18
RMSE for clipped outliers: 1.17
RMSE for removed outliers: 1.30

X_train = X_train_imputed.copy()
X_val = X_val_imputed.copy()
X_test = X_test_imputed.copy()

zero_var_cols = [col for col in X_train.columns if X_train[col].nunique() == 1]
print("Zero variance columns:", zero_var_cols)

Zero variance columns: []

df_corr = pd.concat([X_train, y_train], axis=1)
corr = df_corr.corr()

plt.figure(figsize=(12, 10))
sns.heatmap(corr, annot=True, fmt=".2f", cmap="coolwarm", square=True, center = 0)
plt.title("Correlation Heatmap")
plt.show()

columns_to_drop_manual = ["CO(GT)", "PT08.S1(CO)", "NOx(GT)", "PT08.S3(NOx)", 'PT08.S4(NO2)', 'PT08.S5(O3)']

X_train_manual = X_train.drop(columns=columns_to_drop_manual)
X_val_manual = X_val.drop(columns=columns_to_drop_manual)

print("Liczba kolumn po usunięciu manualnym:", X_train_manual.shape[1])

Liczba kolumn po usunięciu manualnym: 5

from sklearn.decomposition import PCA

pca = PCA(n_components=0.95)
X_train_pca = pca.fit_transform(X_train)
X_val_pca = pca.transform(X_val)

print("Liczba kolumn po PCA:", X_train_pca.shape[1])

Liczba kolumn po PCA: 3

from sklearn.feature_selection import SelectKBest, f_regression

selector = SelectKBest(score_func=f_regression, k=5)
X_train_kbest = selector.fit_transform(X_train, y_train)
X_val_kbest = selector.transform(X_val)
print("Liczba kolumn po SelectKBest:", X_train_kbest.shape[1])

Liczba kolumn po SelectKBest: 5

rmse_orig, rmse_orig_lr = build_and_evaluate(X_train, y_train, X_val, y_val)
rmse_manual, rmse_manual_lr = build_and_evaluate(X_train_manual, y_train, X_val_manual, y_val)
rmse_pca, rmse_pca_lr = build_and_evaluate(X_train_pca, y_train, X_val_pca, y_val)
rmse_kbest, rmse_kbest_lr = build_and_evaluate(X_train_kbest, y_train, X_val_kbest, y_val)

print("XGBoost RMSE (oryginalne dane):", rmse_orig)
print("XGBoost RMSE (usunięte ręcznie):", rmse_manual)
print("XGBoost RMSE (PCA):", rmse_pca)
print("XGBoost RMSE (SelectKBest):", rmse_kbest)

print("\nLinear Regression RMSE (oryginalne dane):", rmse_orig_lr)
print("Linear Regression RMSE (usunięte ręcznie):", rmse_manual_lr)
print("Linear Regression RMSE (PCA):", rmse_pca_lr)
print("Linear Regression RMSE (SelectKBest):", rmse_kbest_lr)

XGBoost RMSE (oryginalne dane): 0.3460492680167805
XGBoost RMSE (usunięte ręcznie): 0.5507019851134495
XGBoost RMSE (PCA): 1.9760168198703294
XGBoost RMSE (SelectKBest): 0.5104304904139009

Linear Regression RMSE (oryginalne dane): 1.1758562616031005
Linear Regression RMSE (usunięte ręcznie): 1.3627617601667226
Linear Regression RMSE (PCA): 2.3743754204362375
Linear Regression RMSE (SelectKBest): 1.322532397359445

train_cleaned = pd.concat([X_train, y_train], axis=1)
train_cleaned.to_csv("data/train_cleaned.csv", index=False)

val_cleaned = pd.concat([X_val, y_val], axis=1)
val_cleaned.to_csv("data/val_cleaned.csv", index=False)

test_cleaned = pd.concat([X_test, y_test], axis=1)
test_cleaned.to_csv("data/test_cleaned.csv", index=False)

import pandas as pd

train = pd.read_csv('data/train_cleaned.csv')
val = pd.read_csv('data/val_cleaned.csv')
test = pd.read_csv('data/test_cleaned.csv')

df = pd.concat([train, val, test], axis=0)
df.head(10)

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
df = pd.DataFrame(scaler.fit_transform(df), columns=df.columns)

from sklearn.cluster import KMeans
import matplotlib.pyplot as plt

sse = []
for i in range(1,11):
    km = KMeans(n_clusters = i, random_state=42)
    km.fit(df)
    sse.append(km.inertia_)

plt.plot(range(1,11), sse)
plt.xlabel("K-clusters")
plt.ylabel("Inertia")
plt.title("The Elbow method")
plt.show()

kmeans = KMeans(n_clusters=3, random_state=42)
labels = kmeans.fit_predict(df)
df['Cluster'] = labels

import seaborn as sns

sns.scatterplot(data = df, x = "C6H6(GT)", y = "CO(GT)", hue = "Cluster", palette = "Set1", alpha = 0.6)
plt.title('KMeans Clustering')
plt.xlabel('C6H6(GT)')
plt.ylabel('CO(GT)')
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s=50, c='black', label='Centroids')
plt.legend()
plt.show()

cluster_counts = df["Cluster"].value_counts().sort_index()
cluster_df = cluster_counts.reset_index()
cluster_df.columns = ['Cluster', 'Count']

plt.figure(figsize=(8, 5))
sns.barplot(data=cluster_df, x='Cluster', y='Count', hue = "Cluster", palette="Set1", legend=False)
plt.title('Liczba próbek w każdym klastrze')
plt.xlabel('Klaster')
plt.ylabel('Liczba')
plt.tight_layout()
plt.show()

import seaborn as sns

features = ['CO(GT)', 'C6H6(GT)', 'NOx(GT)', 'NO2(GT)']

df_melted = df.melt(id_vars='Cluster', value_vars=features, 
                    var_name='Cecha', value_name='Wartość')

plt.figure(figsize=(12, 6))
sns.boxplot(data=df_melted, x='Cecha', y='Wartość', hue='Cluster', palette='Set1')
plt.title('Rozkład cech jakości powietrza w zależności od klastra')
plt.legend(title='Klaster')
plt.tight_layout()
plt.show()

def train_and_evaluate_model(model, model_name):
    if model_name == "LightGBM":
        model = model.train(params, lgb_train, valid_sets=[lgb_train, lgb_val])
    elif model_name == "CatBoost":
        model.fit(X_train_scaled, y_train,
                  eval_set=(X_val_scaled, y_val),
                  use_best_model=True,
                  logging_level='Silent')
    else:
        model.fit(X_train_scaled, y_train)


    y_train_pred = model.predict(X_train_scaled)
    y_val_pred   = model.predict(X_val_scaled)

    def compute_metrics(y_true, y_pred):
        rmse = np.sqrt(mean_squared_error(y_true, y_pred))
        mae  = mean_absolute_error(y_true, y_pred)
        r2   = r2_score(y_true, y_pred)
        return rmse, mae, r2

    rmse_train, mae_train, r2_train = compute_metrics(y_train, y_train_pred)
    rmse_val,   mae_val,   r2_val   = compute_metrics(y_val,   y_val_pred)

    plt.figure(figsize=(8, 6))
    plt.scatter(y_val, y_val_pred, alpha=0.5)
    plt.plot([y_val.min(), y_val.max()], [y_val.min(), y_val.max()], 'r--')
    plt.xlabel('Rzeczywiste wartości')
    plt.ylabel('Przewidywane wartości')
    plt.title(f'{model_name} — Predicted vs. True Values')
    plt.grid(True)
    plt.show()


    print(f"--- {model_name} ---")
    print(f"TRAIN →  RMSE: {rmse_train:.4f} | MAE: {mae_train:.4f} | R²: {r2_train:.4f}")
    print(f"VALID →  RMSE: {rmse_val:.4f} | MAE: {mae_val:.4f} | R²: {r2_val:.4f}")

    return {
        "rmse_train": rmse_train,
        "mae_train":  mae_train,
        "r2_train":   r2_train,
        "rmse_val":   rmse_val,
        "mae_val":    mae_val,
        "r2_val":     r2_val
    }

import pandas as pd
import numpy as np
from xgboost import XGBRegressor
from lightgbm import LGBMRegressor, Dataset as LGBDataset
from catboost import CatBoostRegressor
from sklearn.ensemble import RandomForestRegressor
from sklearn.tree import DecisionTreeRegressor
from sklearn.linear_model import LinearRegression, Ridge, Lasso
from sklearn.metrics import mean_absolute_error
import matplotlib.pyplot as plt
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.metrics import PredictionErrorDisplay
from sklearn.preprocessing import PolynomialFeatures
from sklearn.pipeline import Pipeline
from sklearn.model_selection import GridSearchCV

train_cleaned= pd.read_csv('data/train_cleaned.csv')
val_cleaned   = pd.read_csv('data/val_cleaned.csv')

X_train = train_cleaned.iloc[:, :-1]
y_train = train_cleaned.iloc[:, -1]

X_val = val_cleaned.iloc[:, :-1]
y_val = val_cleaned.iloc[:, -1]

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()

X_train_scaled = scaler.fit_transform(X_train)
X_val_scaled  = scaler.transform(X_val)

rf = RandomForestRegressor(random_state=42)
rmse_rf = train_and_evaluate_model(rf, "Random Forest")

--- Random Forest ---
TRAIN →  RMSE: 0.0649 | MAE: 0.0057 | R²: 0.9999
VALID →  RMSE: 0.1628 | MAE: 0.0201 | R²: 0.9995

xgb = XGBRegressor(random_state=42)
rmse_xgb = train_and_evaluate_model(xgb, "XGBoost")

--- XGBoost ---
TRAIN →  RMSE: 0.0307 | MAE: 0.0233 | R²: 1.0000
VALID →  RMSE: 0.3460 | MAE: 0.0758 | R²: 0.9980

CatBoost=CatBoostRegressor(random_state=42)
rmse_cat = train_and_evaluate_model(CatBoost, "CatBoost")

--- CatBoost ---
TRAIN →  RMSE: 0.2166 | MAE: 0.1521 | R²: 0.9991
VALID →  RMSE: 0.3253 | MAE: 0.1766 | R²: 0.9982

Decision_Tree=DecisionTreeRegressor(random_state=42)
rmse_dt = train_and_evaluate_model(Decision_Tree, "Decision Tree")

--- Decision Tree ---
TRAIN →  RMSE: 0.0000 | MAE: 0.0000 | R²: 1.0000
VALID →  RMSE: 0.1453 | MAE: 0.0190 | R²: 0.9996

LinearRegression=LinearRegression()
rmse_lr = train_and_evaluate_model(LinearRegression, "Linear Regression")

--- Linear Regression ---
TRAIN →  RMSE: 1.0599 | MAE: 0.7641 | R²: 0.9796
VALID →  RMSE: 1.1759 | MAE: 0.8096 | R²: 0.9764

pipeline = Pipeline([
    ("poly", PolynomialFeatures(include_bias=False)),
    ("lr", LinearRegression())
])

param_grid = {
    "poly__degree": [1, 2, 3, 4, 5]
}

grid = GridSearchCV(pipeline, param_grid, scoring="neg_root_mean_squared_error", cv=5)
grid.fit(X_train_scaled, y_train)

best_degree = grid.best_params_["poly__degree"]
print(f"Najlepszy stopień wielomianu: {best_degree}")

Najlepszy stopień wielomianu: 3

poly_pipeline = Pipeline([
    ("poly_features", PolynomialFeatures(degree=3, include_bias=False)),
    ("linear_regression", LinearRegression())
])

rmse_poly = train_and_evaluate_model(poly_pipeline, "Polynomial Regression (deg=3)")

--- Polynomial Regression (deg=3) ---
TRAIN →  RMSE: 0.0291 | MAE: 0.0245 | R²: 1.0000
VALID →  RMSE: 0.0312 | MAE: 0.0267 | R²: 1.0000

Ridge=Ridge(random_state=42)
rmse_ridge = train_and_evaluate_model(Ridge, "Ridge")

--- Ridge ---
TRAIN →  RMSE: 1.0599 | MAE: 0.7636 | R²: 0.9796
VALID →  RMSE: 1.1763 | MAE: 0.8092 | R²: 0.9764

Lasso=Lasso(random_state=42)
rmse_lasso = train_and_evaluate_model(Lasso, "Lasso")

--- Lasso ---
TRAIN →  RMSE: 1.6175 | MAE: 1.0156 | R²: 0.9524
VALID →  RMSE: 1.7872 | MAE: 1.0939 | R²: 0.9455

import pandas as pd

train_cleaned = pd.read_csv('data/train_cleaned.csv')
val_cleaned   = pd.read_csv('data/val_cleaned.csv')

X_train = train_cleaned.iloc[:, :-1]
y_train = train_cleaned.iloc[:, -1]

X_val = val_cleaned.iloc[:, :-1]
y_val = val_cleaned.iloc[:, -1]

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()

X_train_scaled = pd.DataFrame(scaler.fit_transform(X_train), columns=X_train.columns)
X_val_scaled  = pd.DataFrame(scaler.transform(X_val), columns=X_val.columns)

from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
import numpy as np
import matplotlib.pyplot as plt

def feature_improtances(X_train, y_train, X_val, y_val):
    rf = RandomForestRegressor(random_state=42)
    rf.fit(X_train, y_train)
    y_pred = rf.predict(X_val)

    mse = mean_squared_error(y_val, y_pred)
    mae = mean_absolute_error(y_val, y_pred)
    r2 = r2_score(y_val, y_pred)

    feature_importances = rf.feature_importances_
    feature_names = X_train.columns

    importances_df = pd.DataFrame({'Feature': feature_names, 'Importance': feature_importances})

    importances_df = importances_df.sort_values(by='Importance', ascending=False)

    importances_df.plot(kind='bar', x='Feature', y='Importance', legend=False)
    plt.title('Feature Importances')
    plt.xlabel('Importance')
    plt.ylabel('Feature')
    plt.tight_layout()
    plt.show()
    
    print(f'RMSE: {round(np.sqrt(mse), 4)}')
    print(f'MAE: {round(mae, 4)}')
    print(f'R2: {round(r2, 4)}')

feature_improtances(X_train_scaled, y_train, X_val_scaled, y_val)

RMSE: 0.1628
MAE: 0.0201
R2: 0.9995

X_train_fi = X_train_scaled.drop(["PT08.S2(NMHC)"], axis = 1)
X_val_fi = X_val_scaled.drop(["PT08.S2(NMHC)"], axis = 1)

feature_improtances(X_train_fi, y_train, X_val_fi, y_val)

RMSE: 1.4297
MAE: 0.8204
R2: 0.9651

X_train_fi.drop(["CO(GT)"], axis = 1, inplace = True)
X_val_fi.drop(["CO(GT)"], axis = 1, inplace = True)

feature_improtances(X_train_fi, y_train, X_val_fi, y_val)

RMSE: 1.4453
MAE: 0.8142
R2: 0.9644

from sklearn.model_selection import GridSearchCV

param_grid = {
    'n_estimators': [200, 300, 500],          # liczba drzew w lesie
    'max_depth': [None, 10, 20],           # maksymalna głębokość drzewa
    'min_samples_split': [2, 5],               # minimalna liczba próbek do podziału węzła
    'min_samples_leaf': [1, 2],                 # minimalna liczba próbek w liściu
    'max_features': ['sqrt', 'log2', 1.0],
}

rf = RandomForestRegressor(random_state=42)

grid_search = GridSearchCV(
    param_grid=param_grid,
    estimator=rf,
    cv=5,
    scoring='neg_root_mean_squared_error',
    n_jobs=-1,
)

grid_search.fit(X_train_scaled, y_train)


best_model = grid_search.best_estimator_
y_pred = best_model.predict(X_val_scaled)

mse = mean_squared_error(y_val, y_pred)
print("Najlepsze parametry:", grid_search.best_params_)
print("RMSE na zbiorze walidacyjnym:", np.sqrt(mse))
print("RMSE na zbiorze treningowym:", np.sqrt(mean_squared_error(y_train, best_model.predict(X_train_scaled))))

Najlepsze parametry: {'max_depth': 10, 'max_features': 1.0, 'min_samples_leaf': 2, 'min_samples_split': 2, 'n_estimators': 500}
RMSE na zbiorze walidacyjnym: 0.08623124179471915
RMSE na zbiorze treningowym: 0.11464284210695994

from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import PolynomialFeatures
from sklearn.pipeline import Pipeline

poly_pipeline = Pipeline([
    ("poly_features", PolynomialFeatures(degree=3, include_bias=False)),
    ("linear_regression", LinearRegression())
])

poly_pipeline.fit(X_train_scaled, y_train)
y_pred = poly_pipeline.predict(X_val_scaled)
mse = mean_squared_error(y_val, y_pred)

print("RMSE na zbiorze treningowym:", round(np.sqrt(mean_squared_error(y_train, poly_pipeline.predict(X_train_scaled))),4))

best_model = poly_pipeline

print("\nRMSE na zbiorze walidacyjnym:", round(np.sqrt(mse),4))
print("MAE na zbiorze walidacyjnym:", round(mean_absolute_error(y_val, y_pred),4))
print("R2 na zbiorze walidacyjnym:", round(r2_score(y_val, y_pred),4))

RMSE na zbiorze treningowym: 0.0291

RMSE na zbiorze walidacyjnym: 0.0312
MAE na zbiorze walidacyjnym: 0.0267
R2 na zbiorze walidacyjnym: 1.0

test_cleaned = pd.read_csv('data/test_cleaned.csv')

X_test = test_cleaned.iloc[:, :-1]
y_test = test_cleaned.iloc[:, -1]

X_test_scaled = pd.DataFrame(scaler.transform(X_test), columns=X_test.columns)
y_pred = best_model.predict(X_test_scaled)

mse = mean_squared_error(y_test, y_pred)
mae = mean_absolute_error(y_test, y_pred)
r2 = r2_score(y_test, y_pred)

print("RMSE na zbiorze testowym:", round(np.sqrt(mse),4))
print("MAE na zbiorze testowym:", round(mae,4))
print("R2 na zbiorze testowym:", round(r2,4))

RMSE na zbiorze testowym: 0.0319
MAE na zbiorze testowym: 0.027
R2 na zbiorze testowym: 1.0

target = pd.concat([y_train, y_val, y_test], axis=0)
target.describe()

count    8991.000000
mean       10.083105
std         7.449820
min         0.100000
25%         4.400000
50%         8.200000
75%        14.000000
max        63.700000
Name: C6H6(GT), dtype: float64

import joblib

joblib.dump(best_model, 'best_model.pkl')

['best_model.pkl']

Nr	Nazwa zmiennej	Opis
0	Date	Data pomiaru w formacie DD/MM/YYYY
1	Time	Godzina pomiaru w formacie HH.MM.SS
2	True CO (mg/m³)	Rzeczywiste stężenie tlenku węgla (CO) zmierzone przez analizator referencyjny
3	PT08.S1 (tin oxide)	Odpowiedź sensora tlenku cyny (nominalnie celowanego w CO)
4	True NMHC (µg/m³)	Rzeczywiste stężenie węglowodorów niemetanicznych (NMHC) przez analizator referencyjny
5	True Benzene (µg/m³)	Rzeczywiste stężenie benzenu przez analizator referencyjny
6	PT08.S2 (titania)	Odpowiedź sensora tlenku tytanu (nominalnie celowanego w NMHC)
7	True NOx (ppb)	Rzeczywiste stężenie tlenków azotu (NOx) w ppb przez analizator referencyjny
8	PT08.S3 (tungsten oxide)	Odpowiedź sensora tlenku wolframu (nominalnie celowanego w NOx)
9	True NO₂ (µg/m³)	Rzeczywiste stężenie dwutlenku azotu (NO₂) przez analizator referencyjny
10	PT08.S4 (tungsten oxide)	Odpowiedź sensora tlenku wolframu (nominalnie celowanego w NO₂)
11	PT08.S5 (indium oxide)	Odpowiedź sensora tlenku indu (nominalnie celowanego w O₃)
12	Temperature (°C)	Temperatura powietrza w stopniach Celsjusza
13	Relative Humidity (%)	Wilgotność względna powietrza w procentach
14	Absolute Humidity	Wilgotność bezwzględna (AH)

	PT08.S5(O3)	T	RH	AH	Unnamed: 15	Unnamed: 16
0	1268.0	13.6	48.9	0.7578	NaN	NaN
1	972.0	13.3	47.7	0.7255	NaN	NaN
2	1074.0	11.9	54.0	0.7502	NaN	NaN
3	1203.0	11.0	60.0	0.7867	NaN	NaN
4	1110.0	11.2	59.6	0.7888	NaN	NaN

Kolumna	Non-Null	Braki (liczba)	Braki (%)
CO(GT)	7 674	1 683	18,0 %
PT08.S1(CO)	8 991	366	3,9 %
NMHC(GT)	914	8 443	90,2 %
C6H6(GT)	8 991	366	3,9 %
PT08.S2(NMHC)	8 991	366	3,9 %
NOx(GT)	7 718	1 639	17,5 %
PT08.S3(NOx)	8 991	366	3,9 %
NO2(GT)	7 715	1 642	17,6 %
PT08.S4(NO2)	8 991	366	3,9 %
PT08.S5(O3)	8 991	366	3,9 %
T	8 991	366	3,9 %
RH	8 991	366	3,9 %
AH	8 991	366	3,9 %

Zmienna	Średnia	Min	Max	Odch. std.	Wariancja	Skośność	Kurtoza	Q1	Mediana	Q3
PT08.S3(NOx)	963.2975	461.00	1935.00	265.9142	70710.3448	0.8312	0.6530	769.0000	920.0000	1131.0000
NO2(GT)	100.2600	19.00	196.00	31.4938	991.8609	0.0741	–0.1430	78.5000	99.0000	122.0000
C6H6(GT)	10.7711	0.50	39.20	7.4181	55.0287	0.9964	0.7950	4.8000	9.1000	14.8000
CO(GT)	2.3536	0.30	8.10	1.4095	1.9867	1.0643	1.0700	1.3000	2.0000	3.1000
NMHC(GT)	231.0254	7.00	1189.00	208.4619	43456.3686	1.4826	1.9560	77.0000	157.0000	318.5000
PT08.S2(NMHC)	966.1161	448.00	1754.00	266.4246	70982.0446	0.3504	–0.4220	754.0000	944.0000	1142.5000
PT08.S1(CO)	1207.8791	753.00	2040.00	241.8170	58475.4599	0.5644	–0.1030	1017.0000	1172.0000	1380.0000
NOx(GT)	143.5018	12.00	478.00	81.8297	6696.1026	0.8645	0.4560	81.0000	128.0000	187.0000
PT08.S4(NO2)	1600.6203	955.00	2679.00	302.2918	91380.3278	0.6900	0.2730	1369.5000	1556.0000	1783.5000
PT08.S5(O3)	1045.8126	263.00	2359.00	400.1347	160107.7481	0.3372	–0.3510	760.0000	1009.0000	1320.0000
AH	0.8319	0.4023	1.4852	0.1785	0.0319	0.7323	0.8370	0.7189	0.8177	0.9275
T	15.6015	6.3000	30.0000	4.8253	23.2836	0.6252	0.0700	11.9000	15.0000	18.3000
RH	49.0502	14.9000	83.2000	15.2667	233.0735	0.0081	–0.8650	36.7000	49.6000	60.5500

	Date	Time	CO(GT)	PT08.S1(CO)	NMHC(GT)	C6H6(GT)	PT08.S2(NMHC)	NOx(GT)	PT08.S3(NOx)	NO2(GT)	PT08.S4(NO2)	PT08.S5(O3)	T	RH	AH	Unnamed: 15	Unnamed: 16
0	10/03/2004	18.00.00	2.6	1360.0	150.0	11.9	1046.0	166.0	1056.0	113.0	1692.0	1268.0	13.6	48.9	0.7578	NaN	NaN
1	10/03/2004	19.00.00	2.0	1292.0	112.0	9.4	955.0	103.0	1174.0	92.0	1559.0	972.0	13.3	47.7	0.7255	NaN	NaN
2	10/03/2004	20.00.00	2.2	1402.0	88.0	9.0	939.0	131.0	1140.0	114.0	1555.0	1074.0	11.9	54.0	0.7502	NaN	NaN
3	10/03/2004	21.00.00	2.2	1376.0	80.0	9.2	948.0	172.0	1092.0	122.0	1584.0	1203.0	11.0	60.0	0.7867	NaN	NaN
4	10/03/2004	22.00.00	1.6	1272.0	51.0	6.5	836.0	131.0	1205.0	116.0	1490.0	1110.0	11.2	59.6	0.7888	NaN	NaN
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
9466	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
9467	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
9468	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
9469	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
9470	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

	PT08.S1(CO)	C6H6(GT)	PT08.S2(NMHC)	PT08.S3(NOx)	PT08.S4(NO2)	PT08.S5(O3)	T	RH	AH
524	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
525	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
526	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
701	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
702	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
...	...	...	...	...	...	...	...	...	...
8111	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
8112	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
8113	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
8114	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
8777	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

	CO(GT)	PT08.S1(CO)	PT08.S2(NMHC)	NOx(GT)	PT08.S3(NOx)	NO2(GT)	PT08.S4(NO2)	PT08.S5(O3)	T	RH	AH	C6H6(GT)
0	0.64	807.0	658.0	33.0	1113.0	45.0	1328.0	428.0	32.4	26.3	1.2523	3.1
1	2.50	998.0	1119.0	207.0	832.0	120.0	1780.0	1057.0	31.6	24.1	1.0994	14.1
2	1.10	968.0	669.0	67.0	1261.0	81.0	1246.0	538.0	12.3	46.8	0.6686	3.3
3	0.98	925.0	749.0	49.4	826.0	48.4	1482.0	739.0	23.5	58.3	1.6672	4.7
4	1.90	1100.0	1022.0	492.0	752.0	193.0	1101.0	1367.0	12.2	30.0	0.4245	11.2
5	2.80	1171.0	1064.0	160.0	881.0	110.0	1639.0	1097.0	16.0	42.4	0.7674	12.4
6	1.40	1097.0	721.0	183.0	746.0	96.0	1266.0	807.0	21.6	46.8	1.1959	4.2
7	0.70	844.0	538.0	158.0	1164.0	109.0	726.0	479.0	3.7	35.2	0.2833	1.4
8	4.00	1305.0	1188.0	403.6	546.0	124.8	1843.0	1524.0	31.0	33.1	1.4620	16.3
9	1.60	1235.0	828.0	118.0	1055.0	83.0	1527.0	1093.0	12.4	60.0	0.8627	6.4

Model	RMSE Train	MAE Train	R² Train	RMSE Val	MAE Val	R² Val	Overfitting
Polynomial Regression (deg=3)	0.0291	0.0245	1.0000	0.0312	0.0267	1.0000	0.0021
Decision Tree	0.0000	0.0000	1.0000	0.1453	0.0190	0.9996	0.1453
Random Forest	0.0649	0.0057	0.9999	0.1628	0.0201	0.9995	0.0979
CatBoost	0.2166	0.1521	0.9991	0.3253	0.1766	0.9982	0.1087
XGBoost	0.0307	0.0233	1.0000	0.3460	0.0758	0.9980	0.3153
Linear Regression	1.0599	0.7641	0.9796	1.1759	0.8096	0.9764	0.1160
Ridge Regression	1.0599	0.7636	0.9796	1.1763	0.8092	0.9764	0.1164
Lasso Regression	1.6175	1.0156	0.9524	1.7872	1.0939	0.9455	0.1697

Model	RMSE Train	RMSE Val	R² Val	Overfitting
Decision Tree	0.0000	0.1459	0.9996	0.1459
Random Forest	0.0649	0.1628	0.9995	0.0979
CatBoost	0.2166	0.3253	0.9982	0.1087
XGBoost	0.0307	0.3460	0.9980	0.3153

Model	RMSE Train	MAE Train	R² Train	RMSE Val	MAE Val	R² Val	Overfitting*	*Underfitting
Polynomial Regression (deg=3)	0.0291	0.0245	1.0000	0.0312	0.0267	1.0000	0.0021	0.0000
Random Forest	0.0649	0.0057	0.9999	0.1628	0.0201	0.9995	0.0979	0.0000

Zbiór Danych¶

Charakterystyka danych¶

Cel projektu¶

1. Podstawowe wymiary i typy danych¶

2. Zestawienie braków danych¶

Tabela statystyk opisowych dla każdej zmiennej¶

Wnioski z tabeli statystyk opisowych:¶

1. Zróżnicowanie skali i rozrzutu¶

2. Asymetria i „ogonki” rozkładów¶

3. „Spiczastość” rozkładów (kurtoza)¶

4. Kwantyle i mediany¶

PT08.S3(NOx)¶

Histogram¶

Boxplot¶

NO2(GT)¶

Analiza wykresów dla zmiennej NO2(GT)¶

Histogram: NO2(GT)¶

Boxplot: NO2(GT)¶

C6H6(GT)¶

Analiza wykresów dla zmiennej C6H6(GT)¶

Histogram: C6H6(GT)¶

Boxplot: C6H6(GT)¶

CO(GT)¶

Analiza wykresów dla zmiennej CO(GT)¶

Histogram: CO(GT)¶

Boxplot: CO(GT)¶

NMHC(GT)¶

Analiza wykresów dla zmiennej NMHC(GT) (węglowodory niemetanowe)¶

Histogram: NMHC(GT)¶

Boxplot: NMHC(GT)¶

PT08.S2(NMHC)¶

Analiza wykresów dla zmiennej PT08.S2(NMHC) (prawdopodobnie pomiar czujnika NMHC)¶

Histogram: PT08.S2(NMHC)¶

Boxplot: PT08.S2(NMHC)¶

PT08.S1(CO)¶

Analiza wykresów dla zmiennej PT08.S1(CO) (pomiar czujnika CO)¶

Histogram: PT08.S1(CO)¶

Boxplot: PT08.S1(CO)¶

NOx(GT)¶

Analiza wykresów dla zmiennej NOx(GT) (tlenki azotu)¶

Histogram: NOx(GT)¶

Boxplot: NOx(GT)¶

PT08.S4(NO2)¶

Analiza wykresów dla zmiennej PT08.S4(NO2) (pomiar czujnika NO2)¶

Histogram: PT08.S4(NO2)¶

Boxplot: PT08.S4(NO2)¶

PT08.S5(O3)¶

Analiza wykresów dla zmiennej PT08.55(O3) (pomiar ozonu)¶

Histogram: PT08.55(O3)¶

Boxplot: PT08.55(O3)¶

AH¶

Analiza wykresów dla zmiennej AH (wilgotność bezwzględna)¶

Histogram: AH¶

Boxplot: AH¶

T¶

Analiza wykresów dla zmiennej T (temperatura)¶

Histogram: T¶

Boxplot: T¶

RH¶

Analiza wykresów dla zmiennej RH (wilgotność względna)¶

Histogram: RH¶

Boxplot: RH¶

Mapa cieplna (heatmap)¶

Mapa cieplna przedstawiająca współczynniki korelacji Pearsona pomiędzy wszystkimi zmiennymi w zbiorze danych.¶

Kluczowe obserwacje:¶

Macierz wykresów punktowych (scatter matrix)¶

Macierz rozrzutu łączy w sobie w jednym wykresie dwie informacje:¶

1. Przekątne (histogramy)¶

2. Zależności między czujnikami i gazami¶

3. Wpływ temperatury¶

4. Wilgotność¶

5. Ogólne wnioski¶

Wykres czasowy danych pomiarowych¶

Analiza każdej zmiennej z wykresu:¶

1. CO(GT)¶

2. PT08.S1(CO)¶

3. NMHC(GT)¶

4. C6H6(GT)¶

5. PT08.S2(NMHC)¶

6. NOx(GT)¶